x x x v i I N T R O D U C T I O N T O V O L U M E 7
facts to more and more general laws. He argues that a change in fundamental
assumptions requires an intuitive grasp of regularities that can then serve as foun-
dational principles for a new theory. Since foundational principles cannot be
derived directly from experience, Einstein concludes that they can never be proven
true. Even if a theory is shown to be in agreement with all available empirical evi-
dence, there may conceivably be another set of principles consistent with the very
same evidence. Hence, the fundamental principles of a physical theory are conven-
tions in Poincaré’s sense: they are not independent of empirical data, but neither are
they determined by them, a point he makes explicitly in Einstein 1921c (Doc. 52).
Nevertheless, despite these epistemological limitations, Einstein believed that
the theorist’s intuition will lead him to select a single theory as superior to all other
potential candidates. In Einstein 1918j (Doc. 7), he makes clear what he believes
the marks of a superior theory to be: the simplicity and logical strength of its foun-
dations, reflecting the “sublime order” of reality. An important example of this
belief is Einstein’s attitude toward the Newtonian versus the relativistic theory of
gravitation. In the second appendix to Einstein 1917a (Vol. 6, Doc. 42, pp. 84–85),
written in 1920, he considers these as two theories that are almost impossible to dis-
tinguish empirically, although they differ widely in their foundations. Yet, Einstein
stressed frequently that the relativistic theory is preferable since its foundational
principles are theoretically superior to those of Newtonian mechanics. By using
quasi-religious imagery to describe Planck’s “Temple of Science,” Einstein implic-
itly acknowledged that he shared his colleague’s conviction that the unity of sci-
ence reflects a transcendent order of nature. His poetic words in Einstein 1921a
(Doc. 51) evoke the same artistic-scientific-spiritual sensibility.
In Einstein 1921c (Doc. 52), Einstein goes beyond these general considerations
to analyze more specifically the relationship between geometry and physical space
in the light of general relativity. Although he agrees with Poincaré’s conventional-
ism as a general methodological program, he does not accept Poincaré’s conclusion
that Euclidean geometry is immune to empirical refutation. Poincaré had argued
that Euclidean geometry would never be abandoned on empirical grounds, since
the association of geometrical objects with physical objects requires physical
assumptions, and it always would be preferable to change the latter than to work in
a more complicated non-Euclidean geometry (Poincaré 1902, pp. 92–109). Like
Poincaré, Einstein accepts that the connection between geometry and experience is
mediated by physical theories, but he nevertheless draws a different conclusion.
Even though physics, too, is conventional, one is not at liberty to choose any phys-
ical theory in order to fit experience to a fixed geometry. For as Poincaré himself
had pointed out, even if fundamental principles are to a certain degree freely
chosen, the totality of empirical knowledge puts severe restrictions on the plausi-